This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON SMART GRID

A Monitoring Technique for Reversed Power Flow

Detection With High PV Penetration Level
Hashem Mortazavi, Hasan Mehrjerdi, Member, IEEE, Maarouf Saad, Senior Member, IEEE,
Serge Lefebvre, Dalal Asber, and Laurent Lenoir, Member, IEEE

AbstractThe
integration
of
renewable
energy
resources (RESs) in power systems poses many research
challenges. Research shows that the RES output may exceed the
consumed power during the day. Consequently, the direction
of the power flow on distribution lines can be reversed during
some periods. As the voltage regulator is normally designed for
unidirectional power flow, this may cause voltage violations on
the distribution feeder. Therefore, most utilities try to set a penetration level (PL) limit for safe operation. On the other hand,
time varying and unbalanced loading are the main characteristics of distribution systems. Moreover installation of intermittent
and nondispatchable photovoltaic (PV) devices increases the
control problems of distribution system. This paper presents an
impedance-based monitoring method for detection of distribution
system current behavior. It will be shown that by utilizing this
monitoring technique, not only the small variation of PV PL can
be easily detected, but also some fast transients such as the effect
of cloud movement on PV system can be monitored. This monitoring technique employs only local measurements of bus voltages
and line current to measure the apparent impedance seen at
the installation point. The practical application of measured
impedance as a monitoring technique shows its effectiveness for
distribution system monitoring in presence of various PV PL.
Index TermsMonitoring, photovoltaic (PV) penetration,
reverse power flow, smart grid, voltage control.

I. I NTRODUCTION
EREGULATION of power system, smart grid, electric
vehicle, and renewable energy integration pose a lot of
challenges to power systems operation. Although a lot of monitoring and protection devices are installed in transmission
system due to its complexity, traditional distribution systems
suffer lack of monitoring systems due to its high number of
feeders and load points.
Distribution systems have been designed for radial and
unidirectional power flows. In this design the active and
reactive power are transferred from the transmission system

Manuscript received May 21, 2014; revised October 7, 2014 and

December 8, 2014; accepted January 23, 2015. This work was supported by
the Research Institute of Hydro-Quebec, Power Systems, and Mathematics,
Varennes, QC, Canada. Paper no. TSG-00488-2014.
H. Mortazavi and M. Saad are with the Department of Electrical and
Computer Engineering, Quebec University (ETS), Montral, QC H3c1 K3,
Canada (e-mail: hmortazavi1@gmail.com).
H. Mehrjerdi is with the Department of Electrical and Computer
Engineering, Qatar University, Doha 2713, Qatar.
S. Lefebvre, D. Asber, and L. Lenoir are with the Research Institute of
Hydro-Quebec, Power Systems, and Mathematics, Varennes, QC J3X 1S1,
Canada.
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSG.2015.2397887

to consumer. Therefore, many protection and regulation

strategies in distribution network are based on this radial
nature. For example, voltage regulators are designed with the
flow of power from the higher voltage to the lower voltage.
Integration of the renewable energy resource (RES) in distribution systems, however, show that, based on the size and
the installation location, the power flow may reverse [1]. For
high penetration of RES, traditional unidirectional distribution system will change to a bidirectional system which needs
a revision of regulation.
According to [2], the reverse power flow is the main
cause of voltage rise in distribution feeders. Customer
load control is one solution for decreasing reverse power.
Asari and Kobayashi [3] proposed that controlling heat pump
water heaters at customer side will minimize reverse power
flow and voltage violation resulting from high penetration
of photovoltaic (PV). Baran et al. [4] analyze high PV
penetration impacts on distribution system protection and
operation. Using flexible ac transmission system and custom power devices (static VAR compensator (SVC) [5],
static synchronous compensator (STATCOM) [6], distribution STATCOM (DSTATCOM) [7], superconducting magnetic
energy storage (SMES) [8]), utilizing electric vehicle [9], storage system [10][12], and active power curtailment technique
recommended in [13][15] are the most attractive techniques
to mitigate those expected voltage violations.
Monitoring distribution system through smart meters and
monitors is one of the basis of smart grid [16]. Different monitoring system approaches and devices have been proposed to
assess different system conditions. Power quality monitoring
systems have been used for a long time not only for analyzing power quality problems but also for load modeling [17].
Qiang et al. [18] proposed voltage monitoring of microgrid at
number of buses to analyze the power quality and reliability
indexes.
Using local data measurement for protection, control, and
monitoring purpose is an attractive alternate solution to the
electric utilities, due to lower installation costs and simplicity
of operation. On-line impedance measurement based on local
data has been proposed for different applications. The most
important application of impedance measurement is in distance relay protection [19]. Vu et al. [20] proposed using local
measured apparent impedance for voltage stability margin estimation. Cespedes and Jian [21] proposed using online grid
impedance measurement for adaptive control of grid connected
inverters.

c 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
1949-3053 
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
2

IEEE TRANSACTIONS ON SMART GRID

Integration of intermittent RES, like solar and wind into

a radial and unidirectional power flow distribution system,
implies incorporating new monitoring devices that are sensitive
to rapid changes of RES generation and power flow direction.
In this research, an impedance measuring technique is proposed as a monitoring tool in distribution systems. The
performance of impedance seen at number of buses as a monitoring technique is analyzed in presence of RES devices.
Due to unbalance loading of distribution system, it will be
shown that any proposed monitoring technique shall have the
capability for separate three-phases monitoring. The speed and
stability of proposed monitoring technique is tested by simulation of important generation fluctuation of connected PV
system due to clouds movement.
This paper is organized as follows. In Section II, the basic
theory of impedance method for monitoring of distribution
feeder with RES will be established. Section III describes the
IEEE 13 node test feeder used as a case study for simulation.
Section IV presents the simulation results of testing the proposed method for unit and nonunit power factor (pf) operation
of PV inverter, fault conditions, load pf variation, and cloud
movement. The practical capability of method is validated by
analyzing the impact of cloud transients on IEEE 34 node test
system. Finally, the conclusion is presented in Section V.
II. M EASURED I MPEDANCE T HEORY AND ITS
A PPLICATION ON R EVERSE P OWER
F LOW D ETECTION
A. Theory of Impedance Method
Distance relay, typically, is the first choice for transmission line protection. In recent years, it was also proposed as a protection device in distribution system [19], [22].
Distance relays use voltage and current as inputs and calculate the apparent impedance seen at the relay location.
This value is a complex number which shows the apparent impedance seen by V/I calculation. A quadrilateral
characteristic-distance relay is proposed in [22] to distinguish the distribution feeder faults in presence of forward
and reverse power flow due to distributed generation (DG)
integration. Uthitsunthorn and Kulworawanichpong [23] proposed using distance relays for distribution feeder protection
with renewable power plant to reduce protection coordination complexity due to impedance based setting of distance
protection. Although this relay is geared at fast and accurate
detection of faults in power system, it was shown that this
method has a very good capability to measure and analyze
the load conditions [24].
When the amount of active and reactive power injected or
absorbed varies at distribution system nodes, the node voltage and line current will change. To clarify the effect of load
variation and RES integration on apparent impedance seen at
any distribution system node, let consider the simple system
shown in Fig. 1.
Consider that PLi and QLi are the total active and reactive power (line power and losses) flowing down from
node i, Pi and Qi are the active and reactive power of load connected to node i, and Rsi + jXsi is the complex line impedance.

Fig. 1. Simple system schematic, the impedance measuring unit connected

to sending end of line connected to bus i.

Voltage drop (Vi ) along the feeder between nodes i and i + 1;

i = 1, . . . , n, is defined as
(1)
Vi = Vi Vi+1 = Zsi Ii
PLi jQLi
(2)
Ii =
Vi
PLi jQLi
Vi = (Rsi + jXsi )
(3)
Vi




Xsi PLi Rsi QLi
Rsi PLi + Xsi QLi
+
j
.
Vi =
Vi
Vi
(4)
As shown in (4), the voltage drop along the feeder has real
and imaginary parts. If the voltage of bus i + 1 is considered
as reference point, that Vi+1 = |V i+1 |0 (this voltage is along
the positive real axis), then from (1) and (4) the voltage at
node i is obtained by


Rsi PLi + Xsi QLi
Vi = Vi+1 +
Vi


Xsi PLi Rsi QLi
. (5)
+ j
Vi
If the VR and VX are defined as (6) and (7), the voltage at
node i can be rewritten as


Rsi PLi + Xsi QLi
(6)
VR = Vi+1 +
Vi


Xsi PLi Rsi QLi
(7)
VX =
Vi
Vi = VR + jVX .
(8)
To calculate apparent impedance seen from node i
Zi =

Vi
VR + jVX
= P jQ
Li
Li
Ii

Zi =

Vi

(9)

Vi

A=

VR + jVX
2
(PLi + jQLi )
PLi + Q2Li
Vi

P2Li + Q2Li
Zi = A (VR + jVX ) (PLi + jQLi )


Zi = A (VR PLi VX QLi ) + j(PLi V X + VR QLi ) .

(10)
(11)
(12)
(13)

For node i, let Piload and Qiload be the active and reactive power of load, PLiloss and QLiloss be the active and
reactive losses of the line, and PiRES , QDiRES denote the
active and reactive generation of RES at node i, respectively.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MORTAZAVI et al.: MONITORING TECHNIQUE FOR REVERSED POWER FLOW DETECTION WITH HIGH PV PL

If there is no RES connected to that node it is assumed that

PRES = 0, QRES = 0. Therefore, we can define the transmitted power through each line of the distribution network as
follows:
n

PLi = PLiloss + P(i+1) +
PL(i+1)
(14)
QLi = QLiloss + Q(i+1) +

i+1
n


QL(i+1)

(15)

i+1

Pi = Piload PiRES
Qi = Qiload QiRES.

(16)
(17)

Then by substituting (14)(17) in (13), it can also be

expressed as
Zi = F (Rsi , Xsi , PLi , QLi , PiRES , QiRES ).

(18)

From (18), it can be seen that the apparent impedance seen

from node i is a function of line impedance, transmitted active
and reactive power of line and the injected or absorbed power
at the node i and rest of the network. Therefore, the impedance
seen at bus i can be used as a monitoring tool for analyzing
the actual system situation.
B. Apparent Impedance Relationship With Power
Flow Direction
As it was stated earlier, voltage and current are two inputs
of the method. The calculated positive sequence of voltage
and currents are used for positive sequence impedance calculation. In this section, the relationship between the calculated
impedance, magnitude and direction of active and reactive
power flow through the line will be established.
Consider two consecutive buses i and i + 1 in Fig. 1,
Zi is the apparent impedance seen on the i side of the line
between those two consecutive buses; PLi and QLi are the
active and reactive powers of the line flowing from side i to
i + 1, respectively; |Vi | is the amplitude of the voltage at bus i.
Therefore the relationships between the measured R and X
and the power transmitted through the lines are given below

  
Si
PLi + jQLi
=
(19)
Ii =
Vi
Vi
Vi
VR + jVX
Zi =
=
(20)

PLi +jQLi
Ii
Vi


PLi + jQLi
(21)
Zi =
|Vi |2 .
P2Li + Q2Li
The real and the imaginary parts of the measured impedance
shown in (21) are written as


PLi
Ri =
(22)
|Vi |2
P2Li + Q2Li


QLi
Xi =
(23)
|Vi |2 .
P2Li + Q2Li
As it can be seen from (22) and (23), the position of measured apparent impedance in R X plane depends on the

Fig. 2.

Mapping of power flow direction on R X diagram [25].

value and direction of active and reactive power. The apparent

impedance measured at bus i has reverse relationship with
power. The larger the power transferred through the line, the
smaller measured R and X.
The sign of Ri in (22) and Xi in (23) are only related to the
sign (direction) of PLi and QLi , respectively. Then, according
to (14) and (15)

yields
If |PiRES | > PL(i+1) + Piload PLi < 0 (24)

yields
If |QiRES | QL(i+1) + Qiload QLi < 0. (25)
Equations (24) and (25) show the conditions that the RES
generation exceeds the consumption and the final location of
the impedance seen at bus i on R X plane will be at third
quarter. For forward Q and reversed P the impedance location
will be at the second part of R X plain.
Based on the direction of active and reactive power flow the
R X diagram can be divided into four parts, Fig. 2 shows the
mapping of power flow direction on R X diagram. Based on
this figure, the impedance located on R axis represented a unity
pf and for x-axis the pf is zero. For the directions of active
and reactive power are from node i to i + 1, the measured
impedance will be located at the first quarter of R X plane.
C. Impedance Measurement
Distance relay is the best tools for impedance measurement
in power system. Because the positive sequence component
is the only common sequence component in all the types of
faults, the measured impedance at distance protection relays
are always based on the phase positive sequence impedance.
Moreover, the apparent impedance measured by a distance
relay will depend on the current and voltage transformers
connections too.
The measured impedance is calculated by three, one phase
measuring units. The simplest ground distance functions use
only a single phase current and a single phase voltage.
Consider that VAN is the phase A to ground voltage and IA is
the current flows through the phase A conductor, the apparent
impedance seen for phase A is given by [26]
VA
.
(26)
ZA =
IA
The impedance calculated using (26) is the positive
sequence impedance seen of the line for all system operation condition except fault conditions. Phase-ground distance

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
4

Fig. 3.

IEEE TRANSACTIONS ON SMART GRID

Schematic of the IEEE 13 node test feeder.

relay usually employ a zero sequence current compensator

for actual system impedance measurement during the fault
conditions. Based on [27], all the phase-ground faults detection units measure the same impedance value for the nonfault
conditions.
III. C ASE S TUDY
The IEEE 13 node test feeder (Fig. 3) is a very small but
relatively highly loaded 4.16 kV feeders. This provides a good
test for the most common features of distribution analysis
software [28]. As this model has unbalanced spot and distributed load, overhead lines and cables, shunt capacitors, and
a voltage regulator consisting of three single-phase units, it is
a good case for testing the impedance method capabilities. The
simulations have been performed by OpenDSS [29] software
and MATLAB [30]. The OpenDSS COM server interfaced
with the MATLAB program is used for simulation.
Wind, photovoltaic (PV), and biomass are the best choices
for RES integration to power system. For large scale power
production at transmission level, biomass system, and wind
turbine are the most obvious solution. On the other hand,
small scale PV system is more attractive solution for residential installation. For RES integration to distribution system
simulation, PV system is chosen.
This research used the built-in PV system model provided
in OpenDSS. Among two options for modeling high penetration of small-scale rooftop PV or large centralized PV,
the second one was chosen for the simulation of impedance
based monitoring system. Hence to produce various scenarios for simulations, the lumped three-phases balanced PV
model was added at three different locations, one close to
substation source bus 632, one near the feeder midpoint
bus 671, and one at the end of feeder at bus 680. For showing
the effect of PV location on the proposed monitoring technique at each simulation only one location was used for PV
installation.
For simulation of different penetration level (PL),
Hoke et al. [31] showed that the 15% limit for PL as a rule of
thumb for most of feeders is very conservative. They showed
that in two-thirds of simulated cases, the maximum PL is
greater than 90%. In this research, to analyze the full capability
of the monitoring technique, the PV output was increased from

Fig. 4. Active power flows through feeder 650-632 and bus 632 voltage
profile with active tap changer and PV is connected to bus 680.

zero to 1.15 p.u. of model total load (P = 3800 kW) with step
size of 0.01 (115 different PLs). At each step, a power flow
was done by OpenDSS and all voltages, currents, and active
and reactive power of feeder were saved. In this paper, except
the under voltage and over voltage limits, other limiting factors such as feeders current rating, the substation transformer
capacity, power quality issues were neglected. The simulation
has been performed for both unit and nonunit pf PV operation.
As the original model of IEEE 13 node test feeder does not
have a load at bus 680 a 360 kW and 270 kvar load connected
to this bus for simulation. A three-phases 60 kvar capacitor
installed at bus 680 for voltage regulation.
IV. S IMULATION R ESULT AND D ISCUSSION
A. Impact of PV PL on Measured Impedance
To show the impact of PV PL the lumped model of PV is
connected to bus 680. Fig. 4 shows the active power flows
through feeder 650-632 and bus 632 voltage profile. As it
can be seen, by increasing PL the active power flow from
the source to grid decreases and the voltage increases. It is
obvious that by increasing the PL the voltage increases, so
the tap changer located between buses 650 and 632 operates
two times to keep the voltage within the standard boundaries.
Fig. 5 shows the measured R and X for different PLs at
bus 632. As it can be seen by increasing PL, the R and X
increase till PL = 0.58 but after this point although X increases
the R decreases. At PL = 88% the R becomes zero and
after this point the R goes negative corresponding the reversed
active power. The reason for the decreasing R is related to the
relationship between the measured R and total P and Q flows
along the line. Based on (22) by decreasing the P-due to the
increased PV PLthe measured R will decrease too.
As it can be seen from Figs. 4 and 5, at the time of
tap changing the measured impedance shows a little change.
This illustrates that the proposed method is sensitive to small
variations of voltage in the system.
Unlike at bus 632, the measured impedance at bus 680 has
a different trend. As it can be seen from Fig. 6 the measured R and X increases by increasing PV PL, but, when the
active power direction is reversed, the value of R decreases
and becomes negative. This shows that, we have reverse active

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MORTAZAVI et al.: MONITORING TECHNIQUE FOR REVERSED POWER FLOW DETECTION WITH HIGH PV PL

PL=108%
PL=80%

PL=20%

(a)

Fig. 5. Measured R and X versus PL at bus 632-phase B with active tap

changer.
PL=108%

PL=93%

PL=80%

PL=20%

(b)
PL=9%

Fig. 6.

Measured R and X versus PL at bus 680.

power while the reactive power direction does not change

(because the PV only operates at unity pf). These graphs
support the basic theory of proposed method described in
Section II-B.
Fig. 7 shows the measured impedance plotted on R X
plane three buses for PV connected at bus 680 and PL variation
between 0 and 115% of nominal load.
Fig. 7 shows the measured impedance loci of different measurement units. Fig. 7(a) shows that by increasing PL of PV
system the reversed active power occurs at PL = 108% at
bus 632. Meanwhile at bus 671 [Fig. 7(b)] this occurs at
PL = 93%. In addition for bus 680 [Fig. 7(c)] the reversed
active power starts at PL = 9%.
Fig. 7 shows that, depending on measurement unit location and PV equipment location, the value and trend of
measured impedance has a unique and detectable characteristic. Based on (20), for any type of energized feeder with
the voltage and current data available, the proposed monitoring technique can be used. As stated earlier, the apparent
impedance measured at any bus has a direct relationship
with PL and reverse with transferred power. The larger the
power transferred through the line (equal to low PL), the
smaller measured R and X, and vice versa. For example at
bus 632, for PL varying from 20% (equal to high load) to
80% (equal to light load), the impedance locations varies from
point (5.904, 2.955) to (3.16, 13.85) on R X plane.

PL=20%
PL=100%

(c)
Fig. 7. Measured impedance at buses 632, 671, and 680 for PV installed
at bus 680 and PL varies from zero to 115%. (a) Measured impedance at
bus 632-phase A. (b) Measured impedance at bus 671-phase A. (c) Measured
impedance at bus 680-phase A.

B. Impedance Variation Due to Normal Load Deviation

Continuous load variation is the intrinsic characteristic of
distribution system. In the last section, it was shown that the
apparent impedance seen by measuring devices has a distinctive changes due to different PV PLs. Fig. 8 illustrates the
typical normalized load patterns applied to the test feeder.
Fig. 9 shows the measured impedance trajectories for
daily load variation. It is found that the small variation
of load shows its impact on the measured impedance. For
example, the measured impedance for phase A changes
from 6.785 + j2.463  at the lightest load condition (58% of
nominal load) to 3.484 + j1.782  for the maximum

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
6

IEEE TRANSACTIONS ON SMART GRID

PL=88%

PL=108%

PL=111%

Fig. 8.

Normalized load patterns of residential loads.

Fig. 10.

Three-phases measured impedance at bus 632.

PL= 9% for 3 phases

Fig. 9.
Three-phases measured impedance at bus 632 for normal load
variation condition.

Fig. 11. Unbalanced voltage effects on three-phases measured impedance at

bus 680.

daily load. The difference between the measured impedances

are due to the unbalanced loading of distribution system.
C. Analyzing the Impact of Unbalanced Distribution
System on the Measured Impedance
Unbalance operation is an intrinsic characteristic of distribution system. It is anticipated that when the voltage and
current of the three-phases are unbalanced, the resultant measured impedance for each phase is not the same as in the other
phases.
For a PV device installed at bus 680 Fig. 10 shows the
loci of measured impedance for each phase at bus 632. It
is worth mentioning that reversed power starts at different
PL for each phase. While the impedance location for the
phase A and C enters the second quarter of R X plane
(for PL = 108% and 111%, respectively) showing small
reversed active power, at the same PL, phase B shows a higher
reversed active power. This phenomenon is due to the unbalanced operation of IEEE 13 node test feeder. It is concluded
that for distribution system, any proposed technique shall be
capable to monitor each phase separately.
Due to the three-phase balanced loads installed at bus 680,
it is anticipated that the measured impedance in each phase
shall show the same trend. Fig. 11 shows the impedance trends

Fig. 12. Measured impedance at buses 632, 671, and 680 for PV installed
at bus 671.

at bus 680. Although the reversed power occurs at the same

PL = 9%, the loci of measured impedance are not the same in
each phase. Because the feeder 671-680 is fed with unbalanced
voltages at bus 671, small differences can be seen on the three
phases impedance loci at bus 680.
Fig. 12 shows the phase-B impedance trends for the same
condition except for the location of PV system which connected to the bus 671. It can be clearly seen that while the
impedance seen from buses 671 and 632 varies a lot with PL

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MORTAZAVI et al.: MONITORING TECHNIQUE FOR REVERSED POWER FLOW DETECTION WITH HIGH PV PL

Fig. 13. Measured impedance at bus 632-phase A, for three different load
pfs connected to bus 671.

variation, the impedance seen from bus 680 contains a small

variation due to bus 680 voltage changes.
Figs. 1012 show that the measured impedance has clear
reaction for downstream various PV penetrations. Meanwhile,
its reaction to upstream PV variation is limited to feeder
voltage deviation.
D. Impact of Load pf Variation
Distribution system loads profiles vary from line to line.
The load pf varies in accordance with active and reactive
power consumed by the load. Fig. 13 shows the effect of
load pf variation on the measured impedance. For simulation the same condition of Fig. 7 (lumped PV connected to
bus 680) is used, except that the pf of load connected to
bus 671 (the biggest load in the test model) is changing form
the lagging 0.8684 to unity pf and then to a 0.8684 leading one.
As it was expected for each pf, the impedance trend varies.
For unit pf, the reversed active power happens at bigger PL
due to increased load active power. On the other hand, for
pf = 0.8684, although the reversed active power happened at
the same condition of positive pf = 0.8684 but the impedance
trend moves further against the origin. This happens because
the negative pf means the VAr generation. Based on (20), any
power generation along the line will decrease the line current,
resulting the bigger impedance (any generation connected to
line, acts as light load situation).
E. Impact of Nonunit pf Operation of PV Inverter
As it was stated earlier, some DG connection standards
such as IEEE 1547 forbid the nonunit pf operation of distributed generation. Some publications [1], [2], [32] consider
the advantages of using RES inverter capability for voltage
support. To analyze the impact of nonunit pf operation of PVs,
three situations are simulated with inverter capable to operate
at pf = 1, 0.95, and 0.9. For example at pf = 0.9 with the
same size inverter, the maximum active power produced by
PV units is limited to 90% of nominal power but at the same
time the system has the capability to produce reactive power
around 43% of nominal power to support voltage profile. For
this simulation, the PV inverter size does not change.

Fig. 14. Measured impedance at bus 632-phase A for three different PV

inverter output pfs.

For PV installed at bus 680 and PL variation from zero

to 115%, Fig. 14 shows the impact of PV inverter pf variation on the apparent measured impedance trend. For simplicity,
the results of phase A of bus 632 are only shown. It is anticipated that by nonunit operation of PV system and with reactive
power generation, some feeders experience not only reverse
active power but also reverse reactive power too. As shown in
Fig. 14, by decreasing the pf from 1 to 0.95 (with the same
inverter size) the reversed active power happened at bigger PL.
For pf = 0.90, due to the reactive power generation of PV
inverter the impedance trend moves from region 1 to 4 at
PL = 102% (shows the reversed reactive power with forward
active power) and then finally enters to third region at the
PL = 110%.
F. Investigating the Measured Impedance Trends
During Fault
RES integration to distribution feeder will change the feeder
fault current. Depend on the size and location of RES, the
contribution of this generation to fault current varies [33].
At this condition, nondirectional over-current protections will
operates for reverse faults, upstream of the protected zone.
Therefore some utilities [33], [34], at their interconnection
standard of DG to distribution system, recommend to install
a distance relay. For example, Hydro-Quebec does not accept
over current relay as the main protection and recommend
to install a distance relay for conditions such as: outage of
one of DG installed in the feeder, intermittent DG generation
characteristic, and etc.
In this section the impact of fault on the feeder is investigated on the measured impedance trend. In normal operation
of feeder, the measured apparent impedance is the combination of the feeder impedance and the equivalent impedance of
total connected loads. Considering the compensation impact
of capacitor bank on feeder pf, the load impedance normally
is very close to real axis. During the fault condition, the
measured impedance has a measurable decrease [for example at phase A, from (3.966 + j7.12) to (0.08 + j0.2) ]
and it moves toward the origin. The final location of the measured impedance is determined by the type of fault and the

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
8

IEEE TRANSACTIONS ON SMART GRID

Fig. 16.

Cloud transients impact on PV output [35].

Fig. 15. Measured impedance at bus 632 for three-phase to ground fault at
bus 671. The small red circle is a nondirectional impedance relay.

fault impedance. The value of the apparent impedance during

the fault is very close to the actual impedance of feeder
between the measuring point and fault point plus the fault
impedance.
Fig. 15 shows the trend of measured impedance at bus 632
for a three-phases to ground short circuit occurred at bus 671.
A nondirectional impedance relay is defined with the 80% of
lines 632-671 impedance as setting. This protection is shown
by the red circle in Fig. 15. A thorough investigation shows
that the measured fault impedance value is settled down in the
defined protection zone.
Furthermore to substantial decreases of impedance during
the fault, by increasing the PL of RES, the resultant impedance
will be increased. Considering (9), by increasing the RES
penetration, the current flow through the line decreases and
the measured impedance will be increased. According to the
distance relay literature and the result of this section, it is
shown that the fault zone is completely different than normal
operation of measured impedance.
G. Analyzing the Capability of Proposed Method
for Fast Transients
Normally distribution system loads are unbalanced, time
varying [28]. By integration of RES like wind and solar, the
distribution system faces intermittent and fast transients like
in distributed generation.
For PV systems, the solar irradiation energy varies slowly
over time from sun rise to sunset with a preknown patterns
correlated with the longitude and latitude of installation site.
The faster transient involved with solar generation is related
to cloud movements. The PV system generation fluctuation
is function of many factors. The centralized or distributed
PV system, the total installed capacity, local weather pattern and the PV panel type are the main important factors
which determine the transient characteristic of PV system
output [35].
The cloud movement causes a fast transient on PV generation, and is one of the biggest challenges for increasing

Fig. 17. Voltage and active power variation at bus 632 for PV installed at
bus 680 while cloud passing transients.

the PV installation. Basically after a cloud movement over

the installation site, the PV generation decreases dramatically.
Afterward a solar ramp will increase the output of PV system.
For distributed rooftop PV, these types of transients have less
challenge for distribution system operation due to the little
possibilities of simultaneous ramp up and down of all PV
units. The PV plant size, its layout, wind speed and cloud
movement direction are dominant factors affecting the generation and consequently the feeder voltage profile [35]. In
other hand, for PV farm where all the PV panels are installed
in a small area, the impact of cloud transients is substantial
on operation and control of distribution system [36].
In this part of this paper, the impedance method capability will be analyzed by fast transient of cloud movement.
Fig. 16 shows the 2900 s simulation of cloud movement on PV
system generation [35]. The time interval between each sample is one second. As it was stated earlier, among two options
of modeling high penetration of small-scale distributed rooftop
PV or large centralized PV, the second one was chosen for the
simulation of impedance based monitoring system. Therefore,
due to smaller installation area of centralized PV system, the
cloud movement transient impact will be severe.
For the PV system installed at bus 680, Fig. 17 shows the
impact of cloud transients on the active power and voltage of

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MORTAZAVI et al.: MONITORING TECHNIQUE FOR REVERSED POWER FLOW DETECTION WITH HIGH PV PL

Fig. 18.

Cloud transients impact on measured R at bus 632.

Fig. 21. IEEE 34 node test feeder. Nine 200 kW PV unit installed at yellow buses. The location and direction of four impedance measuring units are
shown by red arrow [38].

Fig. 19.

Fig. 20.

Cloud transients impact on measured X at bus 632.

Cloud transients impact on measured impedance at bus 632.

bus 632 (connected to the substation bus). In some cases the

PV system output ramp down from 1 to 0.4 p.u. in less than
15 s (for example check between 1800 to 1820 s at Fig. 16).
In comparison with Fig. 4, these fast transients cause high
fluctuation in power flow and consequently high operation of
tap changers. Due to the time constant of tap changer operation, the consumer electric devices experience some short term
low voltage (when the cloud blocks the sun) and over voltage (when the cloud passes away). Figs. 18 and 19 show the
impact of cloud movement transients on measured R and X at
bus 632. As it can be seen the impedance method can react
to fast transients in distribution system operation due to its
simplicity of calculations.
Fig. 20 shows the measured impedance at bus 632 on the
three-phases for the cloud transients. Although the high spatial
diversity of R and X associated cloud movement, the mapped
R and X on the impedance plane shows a special and detectable
trend for each phase. In the next section, it will be shown that
by defining different operation area on the R X plane this
monitoring technique can be used as a practical tool for better
operation of distribution system.

Fig. 22. Cloud transients impact on measured impedance (for phase A) at

four buses. The red line shows the reverse active power starting point. The
light green line shows the reverse reactive power starting point.

H. Practical Application of Impedance Method

To show a practical application of the impedance based
monitoring system, the IEEE 34 Node Test Feeder is
chosen. This test feeder is an actual feeder located in
Arizona [28][37]. This unbalanced feeder has both spot and
distributed load. For PV integration nine distributed 200 kW
PV units are connected to feeder in different locations as
shown in Fig. 21. Four impedance measuring units are connected to buses 802, 816, 852, and 834. The location and
direction of measuring units are shown in Fig. 21 by the
red arrow.
Considering the worst case operation of PV units, the cloud
movement pattern shown in Fig. 16 is used as the PV unit
sun irradiation input. Fig. 22 shows the apparent impedance
trend measured at those four locations in response to the
cloud movement. For simplicity, only the phase-measured
impedances are shown. While the impedance trend shows the
reversed active power for bus 852, the bus 834 shows the
reversed active and reactive power. This phenomena happens
due to the two capacitors installed at the buses 848 (450 kvar)
and 844 (300 kvar).

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
10

IEEE TRANSACTIONS ON SMART GRID

Fig. 23. Alarm issued for reversed active power at different buses (phase A).

Fig. 24. Alarm issued for reversed active and reactive power at different
buses (phase C).

However, considering the speed and the type of events, some

rules and alarm settings can be defined in the impedancebased method to send alarms or actions to the distribution
system operator or to controlling devices installed at distribution system. For example, in some utilities it is forbidden
for distribution system to feed the transmission system. To
implement this limitation, a flag was defined in the monitoring system to generate an alarm. For normal operation the flag
is zero and for reversed power detection at the measuring point
the flag changes from zero to one and an alarm is issued.
Fig. 23 shows the defined alarm output for reverse power
detection at four measuring points. As it can be seen from
Figs. 22 and 23 at buses 852 and 834 the phase A undergoes some reversed active power due to transient of cloud
passing. Therefore, the alarm is issued repeatedly, whenever reversed active power happens. On the other hand, for
buses 802 and 816, no alarms are issued.
As it was stated earlier, due to two capacitor installed at
buses 848 and 844 for pf correction, power flow analysis shows
the permanent reversed reactive power at bus 834. Therefore,
a flag is defined for reversed reactive power detection. This
limitation is shown in Fig. 22 by light green line.
Fig. 24 shows alarms issued for both reversed active and
reactive power, the result is shown for phase C. Due to unbalanced characteristic of IEEE 34 node, unlike the phase A as
shown in Fig. 22, the phase C undergoes reverse power (both
active and reactive) not only at buses 834 and 852 but also
for buses 802 and 816. For instance, bus 802-phase C shows
reversed active power for short period of times, on the other
hand, endures reverse reactive power for most of the times.
This paper has outlined a monitoring application of apparent impedance measurement. It was shown that the apparent
impedance has considerable capability to use as a monitoring
technique for reverse power flow detection at any condition.
Because, based on (22) and (23), any point (r, x) on the
R X coordinates plane is in one-to-one correspondence with
a point (p, q) in the P Q coordinates plane. Meanwhile, the
application of the method after a substantial changes in the

grid like islanding need to be investigated. After islanding,

the power flow direction in the micro-grid may be changed.
The application of the proposed method for islanded microgrid is the same as protection device application. Therefore,
the optimal location and setting for the measuring unit will be
affected by islanding. Work is continuing in order to put the
method to practical use. The defined alarm and action output
of this monitoring technique can be combined with current
voltage controller of the feeder for better voltage regulation in
presence of RES.
V. C ONCLUSION
In this paper, an impedance-based monitoring technique for
evaluation of current situations of distribution system has been
presented.
The main contribution of this paper is based on measured
impedance that has high capability for detecting different states
of distribution system in presence of various PV PLs.
The results indicate that any proposed monitoring technique
shall have the capability of monitoring for balanced and unbalanced system. The dynamic capability of the proposed method
has been tested by monitoring the fast transient phenomena such as cloud movement on centralized PV units under
unbalanced system conditions of the IEEE 34-bus three-phase
distribution network.
A practical application of the proposed method is simulated
to show the capabilities of measured impedance to monitor
and control of distribution system in presence of various PLs
of PV. Authors are working on using this method for voltage control improvement of distribution system with high PV
integration.
R EFERENCES
[1] P. Jahangiri and D. C. Aliprantis, Distributed Volt/VAr control by PV
inverters, IEEE Trans. Power Syst., vol. 28, no. 3, pp. 34293439,
Aug. 2013.
[2] W. Yang, Z. Peng, L. Wenyuan, X. Weidong, and A. Abdollahi, Online
overvoltage prevention control of photovoltaic generators in microgrids,
IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 20712078, Dec. 2012.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MORTAZAVI et al.: MONITORING TECHNIQUE FOR REVERSED POWER FLOW DETECTION WITH HIGH PV PL

[3] M. Asari and H. Kobayashi, Method of controlling reverse power flow

of PV system with heat pump water heater, in Proc. 3rd IEEE PES
Int. Conf. Exhibit. Innov. Smart Grid Technol. (ISGT Europe), Berlin,
Germany, 2012, pp. 16.
[4] M. E. Baran, H. Hooshyar, S. Zhan, and A. Huang, Accommodating
high PV penetration on distribution feeders, IEEE Trans. Smart Grid,
vol. 3, no. 2, pp. 10391046, Jun. 2012.
[5] N. Daratha, B. Das, and J. Sharma, Coordination between OLTC and
SVC for voltage regulation in unbalanced distribution system distributed
generation, IEEE Trans. Power Syst., vol. 29, no. 1, pp. 289299,
Jan. 2014.
[6] T. Aziz, M. Hossain, T. Saha, and N. Mithulananthan, VAR planning with tuning of STATCOM in a DG integrated industrial system,
IEEE Trans. Power Del., vol. 28, no. 2, pp. 875885, Apr. 2013.
[7] C. S. Chen, C. H. Lin, W. L. Hsieh, C. T. Hsu, and T. T. Ku,
Enhancement of PV penetration with DSTATCOM in Taipower distribution system, IEEE Trans. Power Syst., vol. 28, no. 2, pp. 15601567,
May 2013.
[8] K. Byung-Kwan, K. Seung-Tak, B. Sun-Ho, and P. Jung-Wook, Effect
of a SMES in power distribution network with PV system and
PBEVs, IEEE Trans. Appl. Supercond., vol. 23, no. 3, Jun. 2013,
Art. ID 5700104.
[9] P. Goli and W. Shireen, PV integrated smart charging of PHEVs based
on DC link voltage sensing, IEEE Trans. Smart Grid, vol. 5, no. 3,
pp. 14211428, May 2014.
[10] F. Marra, Y. Guangya, C. Traeholt, J. Ostergaard, and E. Larsen,
A decentralized storage strategy for residential feeders with
photovoltaics, IEEE Trans. Smart Grid, vol. 5, no. 2, pp. 974981,
Mar. 2014.
[11] J. von Appen, T. Stetz, M. Braun, and A. Schmiegel, Local voltage control strategies for PV storage systems in distribution grids, IEEE Trans.
Smart Grid, vol. 5, no. 2, pp. 10021009, Mar. 2014.
[12] H. Taheri, O. Akhrif, and A. F. Okou, Contribution of PV generators
with energy storage to grid frequency and voltage regulation via nonlinear control techniques, in Proc. 39th Annu. Conf. IEEE Ind. Electron.
Soc. (IECON), Vienna, Austria, 2013, pp. 4247.
[13] K. Ogimoto, I. Kaizuka, Y. Ueda, and T. Oozeki, A good fit:
Japans solar power program and prospects for the new power system,
IEEE Power Energy Mag., vol. 11, no. 2, pp. 6574, Mar./Apr. 2013.
[14] J. Appen, M. Braun, T. Stetz, K. Diwold, and D. Geibel, Time in the
sun: The challenge of high PV penetration in the German electric grid,
IEEE Power Energy Mag., vol. 11, no. 2, pp. 5564, Mar./Apr. 2013.
[15] R. Tonkoski, L. A. C. Lopes, and T. H. M. El-Fouly, Coordinated
active power curtailment of grid connected PV inverters for overvoltage
prevention, IEEE Trans. Sustain. Energy, vol. 2, no. 2, pp. 139147,
Apr. 2011.
[16] M. P. Tcheou et al., The compression of electric signal waveforms for
smart grids: State of the art and future trends, IEEE Trans. Smart Grid,
vol. 5, no. 1, pp. 291302, Jan. 2014.
[17] I. F. Visconti, D. A. Lima, J. M. C. de Sousa Costa, and
N. Rabello de B. C. Sobrinho, Measurement-based load modeling using
transfer functions for dynamic simulations, IEEE Trans. Power Syst.,
vol. 29, no. 1, pp. 111120, Jan. 2014.
[18] F. Qiang et al., Microgrid generation capacity design with renewables
and energy storage addressing power quality and surety, IEEE Trans.
Smart Grid, vol. 3, no. 4, pp. 20192027, Dec. 2012.
[19] H. Shateri and S. Jamali, Measured impedance at source node of a distribution feeder for inter phase faults, in Proc. IEEE PES Transmiss.
Distrib. Conf. Expo., Vancouver, BC, Canada, 2010, pp. 16.
[20] K. Vu, M. M. Begovic, D. Novosel, and M. M. Saha, Use of local
measurements to estimate voltage-stability margin, IEEE Trans. Power
Syst., vol. 14, no. 3, pp. 10291035, Aug. 1999.
[21] M. Cespedes and S. Jian, Adaptive control of grid-connected inverters
based on online grid impedance measurements, IEEE Trans. Sustain.
Energy, vol. 5, no. 2, pp. 516523, Apr. 2014.
[22] I. Chilvers, N. Jenkins, and P. Crossley, Distance relaying of 11 kV
circuits to increase the installed capacity of distributed generation,
IEEE Proc. Gener. Transmiss. Distrib., vol. 152, no. 1, pp. 4046,
Jan. 2005.
[23] D. Uthitsunthorn and T. Kulworawanichpong, Distance protection of
a renewable energy plant in electric power distribution systems, in Proc.
Int. Conf. Power System Technol. (POWERCON), Hangzhou, China,
2010, pp. 16.
[24] J. Roberts, A. Guzman, and E. O. Schweitzer, III, Z= V/I does not make
a distance relay, in Proc. Annu. Western Protect. Relay Conf. (WPRC),
Spokane, WA, USA, 1993, pp. 1921.

11

[25] C. R. Mason, The Art and Science of Protective Relaying. Hoboken, NJ,
USA: Wiley, 1956.
[26] Y. Hase, Handbook of Power System Engineering. Hoboken, NJ, USA:
Wiley, 2007.
[27] H. Shateri and S. Jamali, Measured impedance by distance relay elements applied on a distribution feeder in a single phase to ground fault,
in Proc. IEEE/PES Power Syst. Conf. Expo. (PSCE), Seattle, WA, USA,
2009, pp. 17.
[28] W. H. Kersting, Distribution System Modeling and Analysis. Boca Raton,
FL, USA: CRC Press, 2012.
[29] R. C. Dugan, Reference Guide: The Open Distribution System
Simulator (OpenDSS), Elect. Power Res. Inst. Inc., Palo Alto, CA, USA,
2012.
[30] Using Matlab Version 5 Guide, The Mathworks, Inc., Natick, MA, USA,
vol. 5, 1998.
[31] A. Hoke, R. Butler, J. Hambrick, and B. Kroposki, Steady-state analysis of maximum photovoltaic penetration levels on typical distribution
feeders, IEEE Trans. Sustain. Energy, vol. 4, no. 2, pp. 350357,
Apr. 2013.
[32] H. Vahedi, K. Al-Haddad, Y. Ounejjar, and K. Addoweesh, Crossover
switches cell (CSC): A new multilevel inverter topology with maximum
voltage levels and minimum DC sources, in Proc. 39th Annu. Conf.
IEEE Ind. Electron. Soc. (IECON), Vienna, Austria, 2013, pp. 5459.
[33] Requirements for the Interconnection of Distributed Generation to the
Hydro-Qubec Medium-Voltage Distribution System (Between 750 V to
44000 V), Hydro-Qubec Standard E.12-01, 2004.
[34] Distributed Generation Technical Interconnection Requirements
Interconnections at Voltages 50 kv and Below, Rev. 3, Hydro One
Networks Inc. Standard DT-10-015, 2013.
[35] R. C. Dugan and T. E. McDermott, An open source platform for collaborating on smart grid research, in Proc. IEEE Power Energy Soc.
Gen. Meeting, San Diego, CA, USA, 2011, pp. 17.
[36] T. Godfrey et al., Modeling smart grid applications with co-simulation,
in Proc. 1st IEEE Int. Conf. Smart Grid Commun. (SmartGridComm),
Gaithersburg, MD, USA, 2010, pp. 291296.
[37] K. Christakou, J. LeBoudec, M. Paolone, and D. C. Tomozei, Efficient
computation of sensitivity coefficients of node voltages and line currents
in unbalanced radial electrical distribution networks, IEEE Trans. Smart
Grid, vol. 4, no. 2, pp. 741750, Jun. 2013.
[38] S. Shafiee, M. Fotuhi-Firuzabad, and M. Rastegar, Investigating the
impacts of plug-in hybrid electric vehicles on power distribution
systems, IEEE Trans. Smart Grid, vol. 4, no. 3, pp. 13511360,
Sep. 2013.

Hashem Mortazavi received the B.Sc. and M.Sc.

(Hons.) degrees in electrical engineering from the
Ferdowsi University of Mashhad, Mashhad, Iran,
in 1998 and 2001, respectively. He is currently
pursuing the Ph.D. degree in electrical engineering
from Quebec University (cole de Technologie
Suprieure), Montreal, QC, Canada.
His current research interests include smart grid,
renewable energy integration, and generation protection coordination.

Hasan Mehrjerdi (M14) received the B.Sc. and

M.Sc. degrees in electrical engineering from the
Ferdowsi University of Mashhad, Mashhad, Iran,
and Tarbiat Modares University, Tehran, Iran, in
1998 and 2002, respectively, and the Ph.D. degree
in electrical engineering from Quebec University
(cole de Technologie Suprieure), Montreal, QC,
Canada, in 2010.
From 2011 to 2013, he was with the Department
of Power Systems and Mathematics, Research
Institute of Hydro-Quebec, Varennes, QC, Canada.
In 2014, he was at Abengoa Research, Seville, Spain, as a Senior Power
System Researcher. In 2015, he joined Qatar University, Doha, Qatar, as an
Assistant Professor. His current research interests include power system and
control studies, integration of renewable energy resources, and smart grid.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
12

IEEE TRANSACTIONS ON SMART GRID

Maarouf Saad (SM08) received the Bachelors

and Masters degrees from cole Polytechnique of
Montreal, Montreal, QC, Canada, in 1982 and 1984,
respectively, and the Ph.D. degree from McGill
University, Montreal, in 1988, all in electrical
engineering.
He joined cole de Technologie Suprieure,
Montreal, in 1987, where he teaches control theory.
His current research interests include nonlinear control and optimization applied to power systems,
flight control, and robotics.

Dalal Asber received the B.Sc. degree in electrical engineering from the University of Damascus,
Damascus, Syria, in 1979, and the M.Sc. and Ph.D.
degrees in automatic control from Institut National
Polytechnique, Nancy, France, in 1988 and 1991,
respectively.
She was a Post-Doctoral Researcher at cole
de Technologie Suprieure, Montreal, QC, Canada.
In 1993, she joined the Hydro-Quebec Research
Institute, Varennes, QC, where she was involved in
power system modeling, simulation, and control.

Serge Lefebvre received the Bachelors and

Masters degrees in electrical engineering from
cole Polytechnique de Montreal, Montreal, QC,
Canada, and the Ph.D. degree in electrical engineering from Purdue University, West Lafayette,
IN, USA.
He was involved in modeling and analysis
of transmission and distribution systems at the
Research Institute of Hydro-Quebec, Varennes, QC.
His current research interests include energy and distribution management applications and evolution of
the network toward a smart grid.

Laurent Lenoir (M06) received the B.Sc. degree in electrical engineering from cole Suprieure dIngnieurs en lectrotechnique et lectronique
dAmiens, Amiens, France; the M.Sc.A. degree from cole Polytechnique de
Montral, Montreal, QC, Canada; and the Ph.D. degree in electrical engineering from cole de Technologie Suprieure, Montreal, in 2001, 2004, and 2009,
respectively.
He was at Hydro-Quebec Research Institute, Varennes, QC, where he was
involved in power system operation in distribution and transmission networks,
and evolution of the networks toward a smart grid.